Upper Gastrointestinal Bleeding in Cirrhosis: Clinical and Endoscopic Correlations

Gut: first published as 10.1136/gut.17.1.37 on 1 January 1976. Downloaded from

Gut, 1976, 17,37-40

Upper gastrointestinal bleeding in cirrhosis: clinical and endoscopic correlations

J. TERtS,' J. M. BORDAS, C. BRU, F. DIAZ, M. BRUGUERA, AND J. RODES From the Liver Unit, Hospital Clinico y Provincial, University ofBarcelona, Barcelona, Spain

SUMMARY The clinical data of 180 episodes of upper gastrointestinal bleeding in 168 patients with cirrhosis of the liver are examined. The source of bleeding had been determined by early endoscopy in all cases. In men under the age of 50 years, and without symptoms of liver failure, bleeding was due to ruptured gastro-oesophageal varices in 84 % of cases. Severe liver failure was associated with acute lesions ofgastric mucosa in many cases. No presumptive diagnosis ofthe source ofhaemorrhage could be based on the examination of other clinical data (presence of ascites, mode of presentation and pattern ofbleeding, history of ulcer disease, alcoholism, and previous medication).

It is universally accepted that early endoscopy is the Methods most efficient method of diagnosis in upper gastrointestinal haemorrhage (Palmer, 1969; Dupuy et al., The study is based on 193 episodes of upper gastro- 1971; Sugawa et al., 1973; Bordas et al., 1973; intestinal bleeding in 168 patients with cirrhosis. All Hoare, 1975). Endoscopy has shown that the rupture of them had been admitted to an intensive care unit of gastro-oesophageal varices accounts for rather for hepatic diseases with haematesesis and/or http://gut.bmj.com/ less than halfthe number ofepisodes of haemorrhage maelena. The endoscopic examination was performed in cirrhotics. Other sources of bleeding are gastro- with an Olympus GIF Type D during active bleeding duodenal ulcers or acute mucosal lesions, and many or within 48 hours from the end of bleeding, demon- patients bleed from several lesions simultaneously strated by gastric aspiration. The diagnosis of portal (Palmer, 1969; Khodadoost and Glass, 1972; hypertension was based on the presence of gastro- Waldram et al., 1974; Rueff, 1974; Hoare, 1975). oesophageal varices. The diagnosis of cirrhosis was In patients with haemorrhage, radiology cannot based on clinical and biochemical data. In 143 on September 24, 2021 by guest. Protected copyright. normally detect acute ulcers or erosions, and the patients a histological examination of the liver was detected lesion may not be the source of bleeding. made (needle or wedge biopsy, post-mortem These considerations would seem to warrant puncture, or necropsy). permanent endoscopy facilities, at least in special A lesion was presumed to be the source of haemor- units where bleeding patients are often received, and rhage when it was actively bleedingduringendoscopy in the emergency wards of general hospitals. In most or when there were signs of recent haemostasia, such cases, patients with gastrointestinal bleeding are as a clot attached to the lesion (Bordas et al., first seen in centres where endoscopic examinations 1973). The patients were divided into four groups are not performed, so that this procedure is delayed according to the source of bleeding: (a) gastro- and management has to be based on clinical obser- oesophageal varices (GOV), (b) acute mucosal lesions vation. either in the stomach or in the duodenum (AML), In this paper a review is made of the clinical data including complete and incomplete erosions, acute of 168 patients with cirrhosis of the liver and gastro- ulcers, and petechiea, (c) chronic gastric or duodenal intestinal bleeding, submitted to early diagnosis by ulcers (GDU), and (d) multiple lesions bleeding endoscopy. The value of those data as an indication simultaneously. Thirteen patients were excluded of the source of bleeding is examined. from the study: in 11 of these cases the source of haemorrhage could not be identified; one had gastric cancer and one Mallory Weiss syndrome. 'Address for reprints: Dr J. Teres, at above address. Each case was investigated for a history of peptic Received for publication 20 October 1974 ulcer disease, or for one of alcoholism (over 80 g 37 Gut: first published as 10.1136/gut.17.1.37 on 1 January 1976. Downloaded from

38 J. Tere's, J. M. Bordas, C. Bru, F. Diaz, M. Bruguera, and J. Rodts ethanol/day for two years or more), and/or previous 50, GOV were the cause of haemorrhage in 79 of medication with drugs potentially harmful to the cases. gastric mucosa (aspirin, corticosteroids, phenylbutazone). HISTORY OF PEPTIC ULCER The severity of liver disease was graded 1 or 2 The antecedent of peptic ulcer or a history suggesting depending on whether there had been symptoms of this disease was found in 29 of the 180 episodes liver failure (jaundice, ascites, encephalopathy, or investigated, but, of these 29, bleeding from GDU weight loss) before the haemorrhage (Maillard and was found in only 34-5 % (Table 3). This percentage Rueff, 1970; Terds et al., 1974). Patients with none was significantly higher than in patients with no of these symptoms or with only one were graded 1, history of peptic ulcer, although this information and those with two or more were graded 2. was of no clinical value. Moreover, 16 of the 26 Haemorrhage was defined as 'massive' when more patients (61 %) in whom bleeding was due to an ulcer than 300 mi/h of blood was required for six hours or were unaware ofthis condition and had no symptoms. more to maintain a normal blood pressure, and as 'relapsing' when there were repeated episodes of HISTORY OF ALCOHOL OR OF PREVIOUS bleeding after admission. The term 'persistent' MEDI CATION haemorrhage was applied when bleeding lasted for One hundred and one of the patients were chronic more than 60 hours or when more than 3000 ml alcoholics, but the distribution of bleeding lesions blood were transfused, independently of the duration was uniform among alcoholics and non-drinkers (Teres et al., 1974). (Table 3). In 43 cases there was a history of drug ingestion, but no relationship could be found Results between this and the cause of the bleeding (Table 3).

The frequency of lesions responsible for the haemor- SEVERITY OF LIVER DISEASE rhage in the 193 episodes examined is reported in The distribution of the lesions responsible for the Table 1. haemorrhage varied slightly according to the severity of the liver disease (Table 3). Ruptured varices were SEX AND AGE the most frequent cause of bleeding, but AML were http://gut.bmj.com/ Gastro-oesophageal varices were more often the more frequent in grade 2 patients than in those of cause of bleeding in patients under the age of 50 grade 1; 55 5 % of the patients of the latter group years (68 7%) than in older patients (43 9%) bled from varices, and this incidence rose to 84 3 % (p < 0-01), and more often in men (58-8%) than in (24 cases out of 32) among men under the age of women (35-8%) (p < 0-01) (Table 2). In men under 50 years.

PATTERNS OF BLEEDING AFTER ADMISSION Lesion Number % Haematemesis was the first manifestation of on September 24, 2021 by guest. Protected copyright. Gastro-oesophageal varices 91 47-1 haemorrhage in 138 cases, and melaena in 42. The Acute mucosal lesion 43 22-3 mode of presentation of the haemorrhage, haemate- Chronic gastroduodenal ulcer 26 13-5 Multiple sites ofbleeding 20 10-4 mesis or melaena, gave no indication of the source Miscellaneous 2 10 of bleeding. GOV were responsible for the bleeding Unknown 11 5-7 in 63% of the cases of 'massive' haemorrhage, in Total 193 100-0 half of the cases of 'persistent' and 'relapsing' Table 1 Source ofbleeding in 193 episodes ofupper haemorrhage, and in 37 8% qf those in whom gastrointestinal haemorrhage in patients with cirrhosis b!eeding stopped spontaneously. The low incidence

Sex Age(yr) Male Female Under 50 Over 50 (no.) (%) (no.) (%) (no.) ('%) (no.) (%) Gastro-oesophageal varices 61 59-8* 19 35-8* 33 68-7* 47 43-9* Acutemucosallesion 22 216 13 24-6 10 20-9 25 23-3 Chronicgastroduodenalulcer 12 11-8 12 22-6 3 6-3t 21 19-6t Multiple sites ofbleeding 7 6-8 9 17-0t 2 4-1 14 13-2 Total 102 100 53 100 48 100 107 100 Table 2 Bleeding lesion in relation to sex andage in 155 patients with cirrhosis *P < 0-01. tP < 0-05. Gut: first published as 10.1136/gut.17.1.37 on 1 January 1976. Downloaded from

Upper gastrointestinal bleeding in cirrhosis: Clinical and endoscopic correlations 39

History of Previous peptic ulcer 4lcoholism medication Liverfailure Ascites Yes No Yes No Yes No 1 2 Yes No (no.) ( %) (no.) ( %) (no.) ( %) (no.) ( %) (no.) ( %) (no.) (%) (no.) (%) (no.) ( %) (no.) ( %) (no.) (% Gastro-oesophageal varices 9 31 82 54 3 49 48-6 42 51-3 19 44-2 72 52 5 66 55-5 25 41 24 42-1 68 55 Acutemucosallesion 7 24-1 36 23-8 22 21-8 21 26-5 11 25-6 32 23-4 23 19-3 20 32-7 14 24-6 29 23-7 Chronic gastroduodenal ulcer 10 34-5 16 10-6 17 16 8 9 11-4 6 13-9 20 14-6 15 12-6 11 18 10 17-5 16 13-2 Multiplesitesofbleeding 3 10-2 17 11-3 13 12-8 7 8-8 7 16 3 13 9 5 15 12 6 5 8-3 9 15-8 10 8-1 Total 29 100 151 100 101 100 79 100 43 100 137 100 119 100 61 100 57 100 123 100 Table 3 Clinical data in relation to source ofhaemorrhage of peptic ulcer as a source of massive (3 8%) and 1973; Dagradi et al., 1973). Aspirin and other drugs, persistent (6%) haemorrhage is noteworthy. phenylbutazone, and corticosteroids have also been considered to be a cause of upper gastrointestinal Discussion haemorrhage from AML (Valman et al., 1968; Crook et al., 1972; Sugawa et al., 1973). In this series, Upper gastrointestinal bleeding in patients with none of these antecedents provided any useful portal hypertension has been classically related to information; the incidence of gastric lesions bore no ruptured oesophageal varices (Hislop et al., 1966), relationship to a history of previous medication. and portacaval shunt operations have been per- These facts may suggest that in patients with portal formed for many years, even though diagnosis of hypertension, substances which irritate the gastric the bleeding site was only presumptive. The present mucosa can induce bleeding from any pre-existing study confirms what other authors have reported: lesion, including varices. that bleeding in patients with cirrhosis may be due From the observation of gastro-oesophageal to several causes other than varices (Palmer, 1969; reflux in patients with cirrhosis and voluminous Khodadoost and Glass, 1972; Bordas, et al., 1973; ascites, it was supposed that the increased abdominal Rueff, 1974; Waldram et al., 1974; Hoare, 1975). pressure caused by the fluid must favour the rupture http://gut.bmj.com/ For the rational management of bleeding in these of varices (Simpson and Conn, 1968). In the present patients, including their medical and/or surgical series, however, there is no relationship between treatment, the source of the haemorrhage must be variceal haemorrhage and ascites. identified. Early endoscopy has remarkably improved The pattern of haemorrhage differed slightly the rate of accurate diagnosis in gastrointestinal between one group of bleeding lesions and another. haemorrhage and has replaced other diagnostic 'Massive' haemorrhage was usually due to rupture measures, including radiography. Many bleeding varices, whereas ulcers did not normally bleed on September 24, 2021 by guest. Protected copyright. lesions, particularly erosions and acute ulcers, aret massively or persistently. underdiagnosed on barium meal examination (Crook A comparison of the sex and age of patients with et al., 1972), and, conversely, radiology may show the source of bleeding gives an interesting result: more than one possible source of bleeding. Endo- 84% of upper gastrointestinal bleeding in men under scopy should be carried out as soon as possible; if the age of 50 years and without liver failure is due to it is delayed, acute gastric lesions may heal and gastro-oesophageal varices. This finding, not p.- disappear. viously reported, may be of some diagnostichelp and The purpose of this study was to investigate may also answer the question as to how many whether some clinical data might be of diagnostic bleeding patients with portal hypertension were help when early endoscopy is not available. For submitted unnecessarily to portacaval shunts before most of the data the results were negative. Even the endoscopy became a normal clinical practice. history of peptic ulcer was of no help, since only Probably very few, because shunt procedures were one-third of the patients with such a history bled usually carried out on young, well-compensated from the ulcer, whereas two-thirds of the patients patients, who had probably had ruptured varices as with a bleeding ulcer had no previous history. is shown in our study. Alcohol has been considered a cause of haemor- This paper confirms that clinical data are not rhagic gastritis, and several series of patients have reliable in localizing the source of bleeding in upper been reported in whom bleeding acute erosions of the gastrointestinal haemorrhage. Consequently, the gastric mucosa developed after a massive intake of need for a permanent endoscopy service becomes alcohol (Khodadoost and Glass, 1972; Sugawa et al., clear, particularly in general hospitals where many Gut: first published as 10.1136/gut.17.1.37 on 1 January 1976. Downloaded from

40 J. Teres, J. M. Bordas, C. Bru, F. Diaz, M. Bruguera, and J. Rodes

bleeding patients are admitted. When early endo- Hoare, A. M. (1975). Comparative study between endoscopy scopy cannot be performed, very few clinical data and radiology in acute upper gastrointestinal haemorrhage. are of any help. Cirrhotic patients under the age of British Medical Journal, 1, 27-30. Khodadoost, J., and Glass, G. B. J. (1972). Erosive gastritis 50, without liver failure and who bleed massively, and acute gastroduodenal ulcerations as source of upper have a more than 80 % chance of having ruptured gastrointestinal bleeding in liver cirrhosis. Digestion, 7, varices, and Sengstaken-Blakemore intubation or 129-138. intravenous vasopressin should be tried. When Maillard, J. N., and Rueff, B. (1970). Indications et resultats des anastomoses porto-caves chez les malades attients de symptoms of advanced liver disease are present and cirrhoses hepatiques. Revue de Medecine, 17, 1005-1012. bleeding is not massive, AML may be suspected. A Palmer, E. D. (1969). The vigorous diagnostic approach to gastric cooling should then be performed as the first upper-gastrointestinal tract hemorrhage. Journal of the therapeutic measure. American Medical Association, 207, 1477-1480. Rueff, B. (1974). Les causes des saignements digestifs d'origine References haute en cas de cirrhose. Nouvelle Presse Medicale, 3, 179-180. Altemeier, W. A., Fullen, W. D. and McDonough, J. J. Simpson, J. A., and Conn, H. (1968). Role of ascites in (1972). Sepsis and gastrointestinal bleeding. Annals of gastroesophageal reflux with comments on the patho- Surgery, 175, 759-770. genesis of bleeding esophageal varices. Gastroenterology, Atik, M. (1972). Platelet dysfunction in massive gastro- 55, 17-25. intestinal bleeding. Gastroenterology, 62, 719. Sugawa, C., Werner, M. H., Hayes, D. F., Lucas, C. E., and Bordas, J. M., Bru, C., Ter6s, J., Pou, J. M., Acero, D., Walt, A. J. (1973). Early endoscopy. A gude to therapy for Vilar Bonet, J., and Rod6s, J. (1973). La fibrogastroscopia acute hemorrhage in the upper gastrointestinal tract. en las hemorragias digestivas altas. Medicina Clinica, 60, Archives of Surgery, 107, 133-137. 634-638. Teres, J., Visa, J., Bordas, J. M., Mas, A., Diaz, F., and Crook, J. N., Gray, L. W., Nance, F. C., and Cohn, I. (1972). Rod6s, J. (1974). Factores en la valoraci6n del riesgo de la Upper gastrointestinal bleeding. Annals of Surgery, 175, hemorragia digestiva en la cirrosis hepatica. Revista. 771-782. Quirurgica Espanolla, 1, 111-1 17. Dagradi, A. E., Lee, E. R., Bosco, D. L., and Stempien, S. J. Valman, H. B., Parry, D. J., and Coghill, N. F. (1968). (1973). The clinical spectrum of hemorrhagic erosive Lesions associated with gastroduodenal haemorrhage, in gastritis. American Journal of Gastroenterology, 60, 30-46. relation to aspirin intake. British Medical Journal, 4, Dupuy, P., Cazard, G., Rueff, B., and Nahum, H. (1971). 661-663. Corr6lations radio-endoscopiques en periode d'hemorragie Waldram, R., Davis, M., Nunnerley, H., and Williams, R.

digestive haute. Presse Medicale, 79, 2309-2311. (1974). Emergency endoscopy after gastrointestinal http://gut.bmj.com/ Hislop, I. G., Waters, T. E., Kellock, T. D., and Swynnerton, haemorrhage in 50 patients with portal hypertension. B. (1966). The natural history of haemorrhage from British Medical Journal, 4, 94-96. oesophageal varices. Lancet, 1, 945-948. on September 24, 2021 by guest. Protected copyright.